Sag et al., Chapter 4 Complex Feature Values 10/7/04 Michael Mulyar.

Slides:



Advertisements
Similar presentations
The Structure of Sentences Asian 401
Advertisements

 Christel Kemke 2007/08 COMP 4060 Natural Language Processing Feature Structures and Unification.
Chapter 4 Syntax.
Syntactic analysis using Context Free Grammars. Analysis of language Morphological analysis – Chairs, Part Of Speech (POS) tagging – The/DT man/NN left/VBD.
Dr. Abdullah S. Al-Dobaian1 Ch. 2: Phrase Structure Syntactic Structure (basic concepts) Syntactic Structure (basic concepts)  A tree diagram marks constituents.
Properties of X-bar Complements, Adjuncts, & Specifiers.
Grammatical Relations and Lexical Functional Grammar Grammar Formalisms Spring Term 2004.
Statistical NLP: Lecture 3
SYNTAX Introduction to Linguistics. BASIC IDEAS What is a sentence? A string of random words? If it is a sentence, does it have to be meaningful?
MORPHOLOGY - morphemes are the building blocks that make up words.
Natural Language Processing - Feature Structures - Feature Structures and Unification.
1 Words and the Lexicon September 10th 2009 Lecture #3.
Syllabus Text Books Classes Reading Material Assignments Grades Links Forum Text Books עיבוד שפות טבעיות - שיעור עשר Chart Parsing (cont) Features.
Analyzing Features of Grammatical Categories Show my head to the people; it is worth seeing. --Feature structure, to Ivan Sag in a dream.
 Christel Kemke 2007/08 COMP 4060 Natural Language Processing Feature Structures and Unification.
Language, Mind, and Brain by Ewa Dabrowska Chapter 2: Language processing: speed and flexibility.
1 Introduction to Computational Linguistics Eleni Miltsakaki AUTH Fall 2005-Lecture 2.
 2003 CSLI Publications Ling 566 Oct 16, 2007 How the Grammar Works.
1 CSC 594 Topics in AI – Applied Natural Language Processing Fall 2009/ Outline of English Syntax.
CS 4705 Lecture 11 Feature Structures and Unification Parsing.
Phrase Structure The formal means of representing constituency.
Constituency Tests Phrase Structure Rules
Syntax Nuha AlWadaani.
Constituents  Sentence has internal structure  The structures are represented in our mind  Words in a sentence are grouped into units, and these units.
Context Free Grammars Reading: Chap 12-13, Jurafsky & Martin This slide set was adapted from J. Martin, U. Colorado Instructor: Paul Tarau, based on Rada.
Meeting 3 Syntax Constituency, Trees, and Rules
1 Features and Unification Chapter 15 October 2012 Lecture #10.
1 LIN 1310B Introduction to Linguistics Prof: Nikolay Slavkov TA: Qinghua Tang CLASS 14, Feb 27, 2007.
Introduction to English Syntax Level 1 Course Ron Kuzar Department of English Language and Literature University of Haifa Chapter 2 Sentences: From Lexicon.
Introduction to Linguistics
LING 388: Language and Computers Sandiway Fong Lecture 17.
BİL711 Natural Language Processing1 Statistical Parse Disambiguation Problem: –How do we disambiguate among a set of parses of a given sentence? –We want.
ESLG 320 Ch. 12 A little grammar language…. Parts of Speech  Noun: a person/place/thing/idea  Verb: an action or a state of being  Adjective: a word.
1 LIN 1310B Introduction to Linguistics Prof: Nikolay Slavkov TA: Qinghua Tang CLASS 13, Feb 16, 2007.
CSE Introduction to Computational Linguistics Tuesdays, Thursdays 14:30-16:00 – South Ross 101 Fall Semester, 2011 Instructor: Nick Cercone
Writing an ERG mal-rule David Mott IBM Emerging Technology Services.
Today Phrase structure rules, trees Constituents Recursion Conjunction
Head-driven Phrase Structure Grammar (HPSG)
Culture , Language and Communication
Rules, Movement, Ambiguity
1 Principles & Parameters Approach in Linguistics - IV Bibhuti Bhusan Mahapatra.
Section 11.3 Features structures in the Grammar ─ Jin Wang.
Linguistic Theory Lecture 5 Filters. The Structure of the Grammar 1960s (Standard Theory) LexiconPhrase Structure Rules Deep Structure Transformations.
Making it stick together…
 Chapter 8 (Part 2) Transformations Transformational Grammar Engl 424 Hayfa Alhomaid.
 2003 CSLI Publications Ling 566 Oct 20, 2015 How the Grammar Works.
 Chapter 4 Noun Phrases Transformational Grammar Engl 424 Hayfa Alhomaid.
1 Introduction to Computational Linguistics Eleni Miltsakaki AUTH Spring 2006-Lecture 2.
◦ Process of describing the structure of phrases and sentences Chapter 8 - Phrases and sentences: grammar1.
 2003 CSLI Publications Ling 566 Oct 17, 2011 How the Grammar Works.
An Introduction to Semantic Parts of Speech Rajat Kumar Mohanty rkm[AT]cse[DOT]iitb[DOT]ac[DOT]in Centre for Indian Language Technology Department of Computer.
3.3 A More Detailed Look At Transformations Inversion (revised): Move Infl to C. Do Insertion: Insert interrogative do into an empty.
Chapter 11: Parsing with Unification Grammars Heshaam Faili University of Tehran.
X-Bar Theory. The part of the grammar regulating the structure of phrases has come to be known as X'-theory (X’-bar theory'). X-bar theory brings out.
Lec. 10.  In this section we explain which constituents of a sentence are minimally required, and why. We first provide an informal discussion and then.
Week 3. Clauses and Trees English Syntax. Trees and constituency A sentence has a hierarchical structure Constituents can have constituents of their own.
LECT. 11 DR. AMAL ALSAIKHAN Government and Case Theories.
Natural Language Processing Vasile Rus
Lecture 2: Categories and Subcategorisation
Beginning Syntax Linda Thomas
BBI 3212 ENGLISH SYNTAX AND MORPHOLOGY
Chapter Eight Syntax.
Part I: Basics and Constituency
Instructor: Nick Cercone CSEB -
Chapter Eight Syntax.
Introduction to Computational Linguistics
Ling 566 Oct 14, 2008 How the Grammar Works.
Principles and Parameters (I)
Structure of a Lexicon Debasri Chakrabarti 13-May-19.
Presentation transcript:

Sag et al., Chapter 4 Complex Feature Values 10/7/04 Michael Mulyar

Complex Feature Structures Main point of this chapter: Complex feature structures are used to describe the notion of Valence. This greatly simplifies the grammar rules and collapses the distinction between words and phrases into one between complete (SATURATED) constituents and ones which lexically require complementation. Main point of this chapter: Complex feature structures are used to describe the notion of Valence. This greatly simplifies the grammar rules and collapses the distinction between words and phrases into one between complete (SATURATED) constituents and ones which lexically require complementation.

Question Does one need phrase structure rules, given the notion of SATURATION? Does one need phrase structure rules, given the notion of SATURATION? Quantifiers Quantifiers Coordination Coordination Modification Modification

Examples (1) Pat relies on Kim. (2) *Pat relies. (3) The child put the toy on the table. (4) *The child put the toy. The notion of valence—a verb requires complements, as lexically specified. Verbs also require specifiers [Fillmore’s Subject Principle]. The notion of valence—a verb requires complements, as lexically specified. Verbs also require specifiers [Fillmore’s Subject Principle].

Architecture: The Phrase Structure Rules Lexical head-complementation Lexical head-complementation Phrasal head-specification Phrasal head-specification Advantages: specifiers not embedded into complement rules; consistent with analysis of NOM in Ch 1; avoids underspecification (e.g. no grammar rules without phrases).

Implementation: Feature Structure Lists VAL features COMPS and SPR described using feature structures (not atomic). VAL features COMPS and SPR described using feature structures (not atomic). Feature structure lists used, where more than one argument is required. Feature structure lists used, where more than one argument is required. Ordering in feature lists consistent with linear word order. Ordering in feature lists consistent with linear word order.

COMPS Feature Transitive Verb Transitive Verb COMPS HEAD noun SPR+ SPR+ Abbreviated: [ COMPS ]. Abbreviated: [ COMPS ]. Feature structure lists, e.g. put [ COMPS ]. Feature structure lists, e.g. put [ COMPS ]. Big advantage: this approach is flexible, i.e. will work with nouns, adjectives, etc. (Reduces need for categories! More on this later) Big advantage: this approach is flexible, i.e. will work with nouns, adjectives, etc. (Reduces need for categories! More on this later)

Questions (1) How to specify a complement PP? e.g. He put the book on the table vs. *He put the book of the table. (2) What about multiple complement frames?

Head-Complement Rule Phrase Independent! Phrase Independent! Predicts a flat structure for the English VP. Predicts a flat structure for the English VP. phrase  H word 1…n phrase  H word 1…n VAL [COMPS ] VAL [COMPS ] VAL [COMPS ] VAL [COMPS ]

Specifiers Verbs and common nouns require specifiers (more on quantifiers later). Verbs and common nouns require specifiers (more on quantifiers later). Specification implemented phrasally (avoids groupings like [the book] [about Jim]). That is, only constituents with COMPS <> can be used with head- specification. Specification implemented phrasally (avoids groupings like [the book] [about Jim]). That is, only constituents with COMPS <> can be used with head- specification.

Head-Specifier Rule phrase  1 H VALSPR phrase  1 H VALSPR VAL [SPR <>] COMPS <> VAL [SPR <>] COMPS <>

The Valence Principle “Unless the rule says otherwise, the mother’s values for the VAL features (SPR and COMPS) are identical to those of the head daughter.” (p. 106) “Unless the rule says otherwise, the mother’s values for the VAL features (SPR and COMPS) are identical to those of the head daughter.” (p. 106)

Words and Phrases SPR and COMPS features allow to dispense with distinctions between N and NOM; VP and V. SPR and COMPS features allow to dispense with distinctions between N and NOM; VP and V. Names can still be applied, but these are no longer vital, since valence specification collapses the distinction between words and phrases. Names can still be applied, but these are no longer vital, since valence specification collapses the distinction between words and phrases. Non-branching domination no longer needed. Non-branching domination no longer needed.

Applying the system: Case Marking in Pipil (Problem 4) Data in problem 4 shows that Pipil is a verb initial language. (VOS in transitive sentences and VS in intransitive sentences.) Data in problem 4 shows that Pipil is a verb initial language. (VOS in transitive sentences and VS in intransitive sentences.) Nouns require a determiner (ne). Nouns require a determiner (ne). Thus, head-specification only applies to nouns. (Verbs are SPR <>). English style head-complement rule will work so long as case is specified. Thus, head-specification only applies to nouns. (Verbs are SPR <>). English style head-complement rule will work so long as case is specified.

Pipil, cont. From what we can tell, case marking is tripartite (S, A, and P). From what we can tell, case marking is tripartite (S, A, and P). Supposing that case is a lexical feature of the noun (a head feature CASE), we can write the following rule: Supposing that case is a lexical feature of the noun (a head feature CASE), we can write the following rule: phrase  H word 1…n phrase  H word 1…n VAL [COMPS ] VAL [COMPS ] VAL [COMPS ] VAL [COMPS ] 1 HEAD noun CASE (S | P) CASE (S | P)

Agreement (A Lexical Analysis) Ch 3 grammar treated agreement via a rule of grammar (e.g. AGR values for NP subjects and VPs must be identical). Ch 3 grammar treated agreement via a rule of grammar (e.g. AGR values for NP subjects and VPs must be identical). Other heads, however, (adjectives and PPs) are not specified for AGR. (Example 31, p. 107). Other heads, however, (adjectives and PPs) are not specified for AGR. (Example 31, p. 107). Sag proposes a lexical solution: verbs and common nouns only need agree with their specifiers. Sag proposes a lexical solution: verbs and common nouns only need agree with their specifiers.

SHAC (Specifier-Head Agreement Constraint) Applies only to verbs and common nouns Applies only to verbs and common nouns HEAD [AGR 1] HEAD [AGR 1] VAL[SPR ] Can be applied to determiner-noun agreement (more on quantifiers below). Can be applied to determiner-noun agreement (more on quantifiers below).

Extending the formalism: Count and Mass Nouns Sag argues that the count / mass distinction is not entirely semantic. Sag argues that the count / mass distinction is not entirely semantic. All smaller portions of furniture are not furniture? (Not infinitely divisible, though.) All smaller portions of furniture are not furniture? (Not infinitely divisible, though.) Nouns specify for determiners (+COUNT or –COUNT), but are themselves unmarked for the feature COUNT. Nouns specify for determiners (+COUNT or –COUNT), but are themselves unmarked for the feature COUNT. The feature COUNT does not project to the phrasal level (e.g. SPR <> feature is used instead). Sag argues that this is consistent with the observation that no English verb requires only either count or mass nouns. The feature COUNT does not project to the phrasal level (e.g. SPR <> feature is used instead). Sag argues that this is consistent with the observation that no English verb requires only either count or mass nouns.

Quantifiers Quantifiers like much or little are –COUNT; quantifiers like a or every are +COUNT. Quantifiers like much or little are –COUNT; quantifiers like a or every are +COUNT. How does the formalism account for: How does the formalism account for: (1) Much furniture was broken. (2)*Much furniture looks bad. (3) Some furniture looks bad. Is (2) a problem for Sag’s prediction? Justice or peace SPR <>, but occur with some quantifiers?

Coordination: A first approach The coordination rule in Ch 3 only allows coordination of conjuncts with identical head values. This clearly undergenerates. The coordination rule in Ch 3 only allows coordination of conjuncts with identical head values. This clearly undergenerates. Sag’s first suggestion is to coordinate conjuncts with identical Valence. (Rule 55, p. 116). Sag’s first suggestion is to coordinate conjuncts with identical Valence. (Rule 55, p. 116). The rule overgenerates somewhat, as NP and S have equivalent valence but do not combine (Example 56). The rule overgenerates somewhat, as NP and S have equivalent valence but do not combine (Example 56).

Coordination Pos and coordination. Pos and coordination. Ellipsis must be accounted for: Ellipsis must be accounted for: John reads and writes letters. (not generated by head-complementation rule) John reads and sends Bill letters. (not elliptical, as grammar correctly generates!)